The embodiments of the invention provide an image style migration model generation method and a mobile terminal. The image style migration model generation method includes the steps: constructing a style training data pair; establishing a neural network according to the structural sequence: an input layer, a preposition layer, at least one residual error characteristic construction layer, a postposition layer and an output layer in order; constructing a loss function based on the mean square error loss of a training original image and a style image and the mean square error of the depth perception feature; and according to the style training data pair and the loss function, training the neural network to generate an image style migration model. By means of the image style migration model generated by the image style migration model generation method, a high-fidelity image after style migration can be quickly obtained by inputting the image to be processed into the image style migrationmodel, without performing cumbersome iterative operations, so that the consumed time is short, and the processor has a light workload, and space distortion of the style migration image can be avoided.